Cost analysis related to dose-response of spinal manipulative therapy for chronic low back pain: outcomes from a randomized controlled trial.

ORIGINAL ARTICLES COST ANALYSIS RELATED TO DOSE-RESPONSE OF SPINAL MANIPULATIVE THERAPY FOR CHRONIC LOW BACK PAIN: OUTCOMES FROM A RANDOMIZED CONTROLLED TRIAL Darcy A. Vavrek, ND, MS, a Rajiv Sharma, PhD, b and Mitchell Haas, DC, MA c

ABSTRACT Objective: The purpose of this analysis is to report the incremental costs and benefits of different doses of spinal manipulative therapy (SMT) in patients with chronic low back pain (LBP). Methods: We randomized 400 patients with chronic LBP to receive a dose of 0, 6, 12, or 18 sessions of SMT. Participants were scheduled for 18 visits for 6 weeks and received SMT or light massage control from a doctor of chiropractic. Societal costs in the year after study enrollment were estimated using patient reports of health care use and lost productivity. The main health outcomes were the number of pain-free days and disability-free days. Multiple regression was performed on outcomes and log-transformed cost data. Results: Lost productivity accounts for most societal costs of chronic LBP. Cost of treatment and lost productivity ranged from $3398 for 12 SMT sessions to $3815 for 0 SMT sessions with no statistically significant differences between groups. Baseline patient characteristics related to increase in costs were greater age (P = .03), greater disability (P = .01), lower quality-adjusted life year scores (P = .01), and higher costs in the period preceding enrollment (P b .01). Pain-free and disability-free days were greater for all SMT doses compared with control, but only SMT 12 yielded a statistically significant benefit of 22.9 pain-free days (P = .03) and19.8 disability-free days (P = .04). No statistically significant group differences in quality-adjusted life years were noted. Conclusions: A dose of 12 SMT sessions yielded a modest benefit in pain-free and disability-free days. Care of chronic LBP with SMT did not increase the costs of treatment plus lost productivity. (J Manipulative Physiol Ther 2014;37:300-311) Key Indexing Terms: Chiropractic; Manipulation; Spinal; Low Back Pain; Costs and Cost Analysis; Health Care Costs

he estimated incidence of moderate back pain is 10% to 15% of the adult population, with a point prevalence of 15% to 30%. 1–3 Low back pain (LBP) is the fifth most common reason for physician office visits in the United States. 4,5 Among people with LBP who see a complementary and alternative medicine (CAM) practitioner for any reason, approximately 49% see a doctor of chiropractic for LBP. 6,7 The use of CAM, including

T a

Assistant Professor, University of Western States, Portland, OR. Associate Professor of Economics, Portland State University, Portland, OR. c Associate Vice President of Research, University of Western States, Portland, OR. Submit requests for reprints to: Darcy A. Vavrek, ND, MS, Assistant Professor, 2900 NE 132nd Ave, Portland, OR 97230 (e-mail: [email protected]). Paper submitted November 26, 2013; in revised form February 4, 2014; accepted March 4, 2014. 0161-4754 Copyright © 2014 by National University of Health Sciences. http://dx.doi.org/10.1016/j.jmpt.2014.03.002 b

300

chiropractic care, has increased significantly over the past 50 years 2,8 and may be associated with lower treatment costs. 9–11 Back pain is the fourth costliest health and productivity burden for US employers among physical and mental health conditions. 7,12 Biennial expenditures were estimated as high as $35.7 billion for ambulatory services for chronic back pain in 2006 to 2007. 13 Including indirect costs, annual US costs of LBP could range from $84.1 to $624.8 billion. 5,7 A recent study in the United Kingdom found that patients with chronic LBP had annual treatment costs that were twice those of matched controls. 14 Furthermore, episodes of LBP have been associated with increased spending on other health conditions. 15 The use of any kind of patient-initiated treatment is intermittent, and ongoing LBP often persists beyond reception of care. 16 The costs paid by the insurer for a course of care provided by a doctor of chiropractic may be as much as 40% less than care provided by a doctor of medicine. 17 Although patients do not seem to seek

Journal of Manipulative and Physiological Therapeutics Volume 37, Number 5

coordinated care across provider types within LPB episodes, perhaps this integration will improve over time. 18 Few studies have been done evaluating the dose effect of spinal manipulative therapy (SMT). 19–21 The scientific evidence on SMT for the relief of chronic LBP has been well discussed in systematic reviews. 22,23 Previously reported data for our trial showed a sustainable within dose-group treatment effect across multiple dose groups to 52 weeks, and although 12 SMT visits were favored, this was not well distinguished from 6 and 18 visits. 24 The purpose of this analysis is to report the incremental cost of treatment and lost productivity as well as the benefit in terms of pain-free (PFDs) and disability-free days (DFDs).

METHODS Design This secondary analysis used data from a prospective openlabel, randomized controlled trial registered on ClinicalTrials. gov (NCT00376350). This trial is discussed in detail by Haas et al. 24 The study was conducted in Portland, Oregon, between March 2007 and July 2011. A total of 400 participants were randomized to receive a dose of 0, 6, 12, or 18 SMT sessions over 18 treatment visits with a doctor of chiropractic. All participants were assigned 3 treatments per week for 6 weeks. Treating physicians provided care at each visit: SMT or a minimal light massage control to isolate the effect of SMT from the effect of touching the patient therapeutically.

Participants Volunteers with a current episode of chronic LBP 25,26 of mechanical origin lasting 3 months or longer were eligible if they were at least 18 years old, ambulatory, and English literate. 27 Participants were required to have some LBP on at least 30 days of the prior 6 weeks and a minimum score of 25 on the 100-point pain intensity scale described below. They were excluded for contraindications to SMT such as active cancer, spine pathology, inflammatory arthropathies, autoimmune disorders, and anticoagulant conditions. Also excluded were potentially confounding conditions including neurodegenerative diseases, pain radiating below the knee, organic referred pain, and disability compensation. 24 This study was approved by the University of Western States Institutional Review Board.

Intervention The SMT treatments consisted of high-velocity, lowamplitude spinal manipulation of the lumbar spine and transition thoracic regions. 28 The light massage used as the control intervention was gentler and of shorter duration than recommended for therapeutic massage practice. 29,30

Vavrek et al Cost Analysis of SMT for LBP

Outcome Measures The main outcome measures for this analysis were PFDs and DFDs, defined as the estimated number of days in the year after randomization that the participant was free of LBP and disability. Both PFDs and DFDs were computed using methods similar to those used by Lave et al 31 to calculate depression-free days and by Dickinson et al 32 to calculate pain-related DFDs. Patients with Modified Von Korff scores of higher than 80 were assumed to be impacted on all intervening days by pain/disability, and those with scores no greater than 20 were assumed free of pain/disability on those days. For those with intermediate pain or disability scores, the number of days of discomfort was assumed to increase linearly with the score. The mean of 2 consecutive scores was used to estimate the number of days with pain/ disability in the intervening time interval. Modified Von Korff LBP pain intensity and pain-related functional disability range from 0 to 100, with higher scores denoting greater severity. 33 The pain score is an average of 3 scales, 0 to 10 each, multiplied by 10: back pain today, worst back pain in the last 4 weeks, and average back pain in the last 4 weeks. Similarly, the disability score is the rescaled average of 3 questions, also 0 to 10 each, covering interference with daily activities, social and recreational activities, and the ability to work outside or around the house. Pain and disability are recognized as key indicators of severity in pain conditions, and indices measuring pain and disability from a composite of several patient responses have desirable psychometric properties for assessments of health. 34,35 Quality-adjusted life years (QALYs) were computed from EuroQol-5D measures collected at baseline and at 12, 24, 39, and 52 weeks. 36,37 Other patient measures including reports of health care use were collected at baseline and at 6, 12, 18, 24, 39, and 52 weeks, summarizing experiences over the prior 4 weeks.

Costs Our objective was to estimate costs from a societal perspective consisting of costs of study-provided care, patient reports of outside care, and lost productivity. Although treatment and lost productivity likely account for most costs related to LBP, some costs were not included in our data. These include the cost transportation and travel time to obtain care, as well as the cost of any equipment purchases, or modifications to accommodate LBP. Consequently, we may underestimate the full societal cost of LBP. Treatment Costs. To estimate costs of treatment not included as part of the study protocol, we used patient reports of outside care use in the previous 4 weeks. Patient reports of care are widely used to estimate use and are regarded as reliable over short periods. 38 Medicare's 2009 national nonfacility (ie, nonhospital) payments were used to estimate costs of patient-reported visits to health care providers. 39 The resource-based relative value units (RVUs) underlying these payments is designed to reflect the resources used to provide

301

302


Journal of Manipulative and Physiological Therapeutics June 2014

Table 1. Estimated Costs Calculated From RVUs CPT Code

Description

99213 Office/outpatient visit with an established patient, the most commonly reported physician service (primary care) 98941 Chiropractic manipulation 97124 Massage therapy 99245 Surgeon/neurologist—office consultation 90806 Psychologist/psychiatrist—individual therapy in office, 45-50 min Naturopath/Homeopath, physical and occupational therapist, acupuncturist, nurse practitioner, and “other health care provider” 99283 Emergency department visit

Estimated Cost $62.35

$33.90 $22.72 $226.50 $87.64 Same as DC visit

$190.59

RVUs, resource-based value units.

services. 40 For example, the cost of each primary care provider visit was $62.35, estimated using Current Procedural Terminology (CPT) code 99213 (office/outpatient visit with an established patient) carrying a physician effort RVU of 0.92, practice expense RVU of 0.77, malpractice insurance RVU of 0.03, and then applying the 2009 physician payment conversion factor of $36.07. Assigned costs of treatment are listed in Table 1. Physician fees for emergency department visits were estimated using Medicare's 2009 national facility payment to account for hospital facility fees under the hospital outpatient prospective payment system. 41 We assumed that the light massage control was minimally therapeutic, being similar to a massage that might be delivered for a few minutes by a friend, and assigned it no cost. 24 Patients reported no hospitalizations, and a single patient from the SMT 6 group reported 2 surgeries. It is unknown whether these were performed inside or outside the hospital. The costs for these 2 surgeries were excluded from the analysis. The minimum cost of back surgery is estimated at $15 000, and including these costs would create an extreme outlier. Costs of patient-reported use of over-the-counter and prescription medicines were estimated using per use rates of $0.20 and $1.00, respectively. Lost Productivity Costs. Lost productivity was computed from patient reports of days with inability to perform usual activities including employment, household work, and selfcare. Work status (and the related concept of productivity) is determined by a complex array of psychological, social, and economic factors. 42 Measurements of lost productivity due to impaired health including pain have mostly relied on direct patient reports of missed work or impaired ability. 43,44 There is a wide variation in the literature regarding methods used in the computation of lost productivity due to impaired health. 7,44,45 Taking this into account, the cost of productivity lost due to LBP was computed from the number of days the patients were kept from their usual activities by LBP. Taking a societal (as opposed to purely employment-based) perspective on productivity, we assumed that on each day that a patient was kept from usual activities, he/she lost 6

hours of productive time (half of an approximate workday involving 8 hours of paid labor and 2 hours of self-care/ household work). Each lost hour was assigned a cost of $15.95, the US median wage in 2009. 46 Annual costs reflect imputed costs incurred in the entire 1-year study period from baseline to 52 weeks.

Statistical Analysis Regression models of the natural logarithm of costs were used to compute the adjusted effect of treatment dose on treatment costs alone, as well as including the cost of lost productivity. Semilog models such as these are commonly used to address skewness in cost data. 47 They yield an adjusted ratio of treatment group to control group costs. Duan's smearing estimator (assuming homoscedasticity) was appropriate for use in all retransformations, as we found no evidence of heteroscedasticity in the cost regressions. 48 Costs of study-provided treatment were excluded because they are additive and nonstochastic. Including these in the multiplicative log models would have created a functional misspecification resulting in biased results. 49 Baseline explanatory variables (covariates) included in all regression models were age, sex, white non-Hispanic, married, college educated, low income (b$20 000), smoker, prior treatment experience with SMT and massage treatment, differential confidence in treatment success between SMT and massage, and baseline pain, disability, and QALY scores. Log of treatment costs in the 4 weeks before baseline was included in treatment cost regressions, whereas log of treatment plus lost productivity in the 4 weeks before baseline was included in treatment plus lost productivity cost regressions. To further demonstrate the variation in costs of treatment and lost productivity, we created 3 illustrative patient profiles by selecting 3 sets of patient characteristics and fitting our regression models post hoc. The patient in profile M has baseline characteristics close to the mean for those enrolled in this study. Profiles L and H depict patients with baseline characteristics that are predictive of lower and higher costs, respectively. The values of the baseline continuous variables in profiles L and H were approximately 1 SD from the values for profile M. The costs of study-provided treatment are included, and the incremental effects of dose as well as some other key predictors of costs in the regression models were computed. 32,50 Linear regression models were used to compute the adjusted effect of treatment dose on the number of PFDs and DFDs experienced by patients. The same covariates used in the cost analysis were included with the exception of baseline cost variables. Statistical testing for all variables was set at a 2-sided α of .05. A full intention-to-treat analysis was conducted with each participant included in the original allocation groups;



Phone Screen (n = 3353) Baseline Exam 1 (n = 522) Baseline Exam 2 (n = 403) Randomization (n = 400)

SMT none (n = 100)

follow-up 6 wk: n = 95 12 wk: n = 85 18 wk: n = 86 24 wk: n = 82 39 wk: n = 84 52 wk: n = 81 Lost to follow-up Address = 1 Medical = 0 Personal = 3 Unknown = 1

SMT 6 (n = 100)


SMT 12 (n = 100)


SMT 18 (n = 100)


Included in analysis n = 95




Reason for exclusion: No follow-up data = 5 No income data = 1

Reason for exclusion: No follow-up data = 1 2 surgeries = 1

Reason for exclusion: No follow-up data = 3 No income data = 1

Reason for exclusion: NA

Fig 1. Study flowchart. SMT, spinal manipulation therapy. (Color version of figure is available online.) missing data were imputed when possible. Two participants declined to give annual income information and so were removed from modeling. Multiple-testing adjustments were not made because this was a secondary analysis. Analyses were conducted with Stata 11.0 (StataCorp, College Station, TX) and SAS 9.2 (SAS Institute, Inc, Cary, NC).

Korff (MVK) scale and mean disability of 45.3 with an average QALY of 0.71. Mean treatment costs in the 4 weeks preceding the baseline were a modest $10.1, on average, and mean costs including the cost of lost productivity in the 4 weeks preceding baseline were substantially larger, $558, on average. No meaningful differences between groups at baseline were noted.

RESULTS

Treatment and Lost Productivity Costs During the Study Period

Figure 1 shows compliance with follow-up. Details are published elsewhere. 24 Compliance was excellent ranging from 81% to 99% per group per time point. There were 9 participants who were entirely lost to follow-up after enrollment and were not included in the analysis. For the cost comparisons, the surgery patient was removed. Two participants refused to report income and were removed from adjusted analyses.

Table 3 shows that unadjusted 1-year costs for treatment outside the study protocol for the groups ranged from $287 to $623. Mean costs for the 1-year study period were much larger when the costs of lost productivity were added to the costs of outside care yielding $2838 to $3815. No statistically significant differences were observed. The adjusted group effects in semilog cost regressions are multiplicative, which are interpreted as ratios of costs compared with control. Annual treatment costs comparisons revealed ratios of 1.15, 1.18, and 0.78, respectively. Costs with lost productivity added resulted in ratios of 0.91, 1.02, and 0.90, respectively. These results were not statistically significant, and no trends across study groups can be inferred given the variability in the costs. In Table 4, the full regression model for the natural log of annual costs shows that greater age (P = .02) and

Baseline Patient Characteristics and Costs Mean age for all groups was 41 years, 15% were minorities, and the groups were divided nearly evenly between men and women (Table 2). More than half of all patients reported having college degrees, 36% were married, and 11% were smokers. Participants had moderate LBP with a mean pain of 51.5 on a 100-point Modified Von

303

304



Table 2. Baseline Patient Characteristics Variable Demographic Age (y) Male Non-white or Hispanic Married College degree Annual income b$20 000 Smoker Confidence in treatment success (0 extremely doubtful–6 extremely certain) Spinal manipulation (6-point Likert scale) Light massage (6-point Likert scale) Previous treatment experience Spinal manipulation Massage Baseline health Pain (100-point MVK scale) Disability (100-point MVK scale) QALY score (EQ-5D) Costs (4 wk preceding baseline) Treatment ($) Treatment + cost of lost productivity ($)

SMT 0

SMT 6

SMT 12

SMT 18

40.9 (14.1) 51% 14% 37% 58% 31% 17%

41.4 (14.8) 51% 18% 28% 63% 27% 13%

41.8 (14.0) 51% 11% 41% 51% 19% 6%

41.2 (13.8) 48% 16% 36% 53% 28% 8%

3.6 (1.2) 3.4 (1.2)

3.8 (1.1) 3.5 (1.2)

3.7 (1.2) 3.4 (1.2)

3.8 (1.1) 3.5 (1.2)

71% 52%

70% 56%

74% 43%

72% 54%

52.2 (16.3) 45.2 (21.8) 0.76 (0.11)

51.0 (18.2) 44.8 (24.0) 0.74 (0.11)

51.6 (17.5) 46.1 (23.4) 0.76 (0.09)

51.5 (16.8) 45.2 (21.8) 0.77 (0.07)

5.5 (13.5) 599.0 (645.8)

15.2 (69.2) 546.1 (591.0)

8.0 (28.0) 553.1 (603.5)

11.7 (34.7) 533.7 (581.4)

MVK, 100-point Modified Von Korff scale; QALY, Quality-adjusted life years; SMT, spinal manipulation therapy. Values are presented as mean (SD), unless otherwise indicated.

Table 3. Cost Comparisons for Treatment and Lost Productivity a Adjusted Cost Ratio b (95% Confidence Interval)

Observed Unadjusted Mean (SD) SMT 0

SMT 6

SMT 12

SMT 18

Treatment costs in preceding 4 wk ($) 0 wk 6 (14) 15 (69) 8 (28) 12 (35) 6 wk 4 (20) 6 (21) 4 (16) 10 (74) 12 wk 23 (47) 52 (152) 13 (29) 32 (102) 18 wk 17 (42) 33 (111) 18 (50) 16 (66) 24 wk 21 (38) 44 (147) 37 (100) 25 (57) 39 wk 32 (72) 62 (170) 36 (87) 26 (50) 52 wk 34 (71) 49 (103) 36 (111) 25 (84) Total 128 (166) 255 (516) 134 (223) 113 (162) Estimated total annual cost of treatment c 315 (419) 623 (1299) 361 (631) 287 (390) Cost of treatment plus lost productivity in preceding 4 wk ($) 0 wk 599 (646) 546 (591) 553 (604) 534 (581) 6 wk 198 (444) 125 (241) 111 (217) 103 (331) 12 wk 312 (517) 266 (462) 181 (270) 221 (350) 18 wk 164 (413) 215 (429) 164 (361) 185 (422) 24 wk 328 (581) 270 (512) 271 (397) 248 (351) 39 wk 240 (414) 318 (515) 233 (412) 220 (395) 52 wk 320 (531) 352 (612) 185 (330) 221 (427) Total 1594 (2312) 1453 (1813) 1169 (1357) 1109 (1282) Estimated total annual cost of 3815 (5384) 3583 (4511) 2991 (3544) 2838 (3282) treatment plus lost productivity c

SMT 6 vs SMT 0

SMT 12 vs SMT 0

SMT 18 vs SMT 0

1.08 (0.81-1.45) 0.94 (0.58-1.52) 1.01 (0.65-1.58) 1.06 (0.65-1.72) 1.16 (0.69-1.95) 1.01 (0.60-1.68) 1.19 (0.67-2.11)

0.93 (0.69-1.25) 0.85 (0.52-1.39) 1.00 (0.64-1.57) 1.16 (0.70-1.90) 1.34 (0.79-2.28) 1.11 (0.66-187) 1.12 (0.63-2.01)

0.88 (0.65-1.17) 0.77 (0.47-1.25) 0.66 (0.42-1.03) 0.78 (0.48-1.27) 0.94 (0.56-1.59) 0.68 (0.41-1.13) 0.75 (0.42-1.33)

1.15 (0.63-2.11)

1.18 (0.64-2.18) 0.78 (0.43-1.43)

0.99 (0.50-1.95) 0.81 (0.40-1.63) 0.99 (0.50-1.96) 0.89 (0.44-1.79) 0.97 (0.48-1.96) 0.62 (0.31-1.27) 0.87 (0.51-1.50) 0.91 (0.57-1. 45)

0.54 (0.27-1.07) 0.62 (0.31-1.26) 0.99 (0.49-2.00) 1.74 (0.85-3.53) 0.91 (0.44-1.85) 0.50 (0.24-1.03) 0.91 (0.52-1.59) 1.02 (0.63-1.64)

0.46 (0.24-0.91) 0.61 (0.30-1.23) 0.78 (0.39-1.56) 1.37 (0.68-2.75) 0.69 (0.34-1.39) 0.55 (0.27-1.11) 0.64 (0.37-1.10) 0.90 (0.56-1.44)

SMT, spinal manipulation therapy. a Excludes cost of study-provided treatment. Costs of study-related SMT over the entire year were as follows: $203 for SMT 6, $407 for SMT 12, and $ 610 for SMT 18. b The adjusted cost ratio is the geometric mean cost for the treatment group divided by the geometric mean costs for the control group adjusted for group differences in baseline covariates. (Adjusted cost ratio − 1) × 100% equals the percentage increase or decrease of treatment group costs compared with the control group (SMT 0) costs. For example, in the last line of the table, annual costs of treatment plus lost productivity were 0.91 times that for SMT 6, 1.02 times that for SMT 12, and 0.90 times that for SMT 18 compared with patients in the SMT 0 group. c Includes imputed costs for weeks not covered by patient reports.



Table 4. Regression Models for LBP-Related PFDs, DFDs and Costs a

PFDs (n = 389; R2 = 0.34)

Variable Intercept Intervention (reference category—SMT 0) SMT 6 SMT 12 SMT 18 Sociodemographic Age Male White non-Hispanic Married College-educated Low income (b$20 to 000) Smoker Previous treatment experience Spinal manipulation Massage Confidence in treatment SMT vs massage (baseline) Baseline health Pain (100-point scale) Disability (100-point scale) QALY (EQ-5D to range −0.109 to 1 Costs (4 wk prior to enrollment) Treatment Treatment and lost productivity

Regression coefficient (95% CI)

P

DFDs (n = 389; R2 = 0.38)

Log Annual Cost of Treatment (n = 388; R2 = 0.40)

Log Annual Cost of Treatment and Lost Productivity (n = 388; R2 = 0.44)




P

326.5 (241.5 to 411.6) b.001 304.6 (231.4 to 377.9) b.001

6.2 (4.3 to 8.1)

P b.001

6.0 (3.5 to 8.5)

P b.001

16.8 (− 3.9 to 37.6) 22.9 (1.8 to 44.0) 18.4 (− 2.3 to 39.1)

.111 .033 .081

12.8 (− 5.1 to 30.6) 19.4 (1.3 to 37.5) 12.3 (− 5.5 to 30.1)

.160 − 0.1 (− 0.6 to 0.4) .036 0.0 (− 0.5 to 0.5) .174 − 0.1 (− 0.6 to 0.4)

.683 0.1 (− 0.5 to 0.7) .936 0.2 (− 0.5 to 0.8) .668 − 0.2 (− 0.9 to 0.4)

.649 .605 .428

− 0.2 (− 0.7 to 0.4) 5.8 (9.4 to 21.0) − 11.6 (− 32.6 to 9.5) − 4.4 (− 21.1 to 12.3) − 5.9 (− 21.2 to 9.4) − 1.7 (− 20.4 to 16.9)

.513 .452 .281 .605 .447 .855

− 0.2 (− 0.7 to 0.3) 0.7 (− 12.4 to 13.8) -3.5 (− 21.6 to 14.6) 2.2 (− 12.2 to 16.6) 1.5 (− 11.6 to 14.7) 1.3 (− 14.7 to 17.4)

.367 0.0 (0.0 to 0.0) .918 − 0.2 (− 0.6 to 0.1) .705 0.3 (− 0.2 to 0.8) .761 0.1 (− 0.2 to 0.5) .819 0.3 (0.1 to 0.6) .872 0.2 (− 0.2 to 0.6)

.020 0.0 (0.0 to 0.0) .207 − 0.4 (− 0.8 to 0.1) .196 0.0 (− 0.6 to 0.6) .467 0.2 (−0.3 to 0.7) .133 0.6 (− 0.1 to 1.0) .370 − 0.1 (− 0.6 to 0.5)

.611 .086 .983 .362 .016 .766

− 18.0 (− 42.2 to 6.2)

.145

3.1 (− 23.9 to 17.7)

.770

0.1 (− 0.5 to 0.6)

.786

0.0 (− 0.7 to 0.7)

.975

12.9 (− 5.5 to 31.3) − 3.6 (− 20.0 to 12.9)

.168 .669

− 2.2 (− 18.0 to 13.6) 2.8 (− 11.4 to 17.0)

.785 .698

0.1 (− 0.3 to 0.5) 0.2 (− 0.1 to 0.6)

.608 .228

0.3 (− 0.3 to 0.8) 0.5 (0.0 to 1.0)

.302 .036

0.6 (− 9.0 to 10.2)

.900

3.5 (− 4.7 to 11.7)

.404

0.1 (− 0.2 to 0.3)

.631

0.0 (− 0.2 to 0.3)

.736

− 2.7 (− 3.3 to − 2.0) − 0.0 (− 0.5 to 0.5)

b.001 .943

− 0.5 (− 1.0 to 0.1) − 1.4 (− 1.8 to − 1.0)

.083 b.001

0.0 (0.0 to 0.0) 0.0 (0.0 to 0.0)

.125 .005

0.0 (0.0 to 0.0) 0.0 (0.0 to 0.0)

.206 .102

.011 104.5 (33.5 to 175.5)

.004

106.9 (24.5 to 189.4)

NA NA

NA NA

− 3.3 (− 5.1 to − 1.4)

0.4 (0.3 to 0.5) NA

.001

b.001

− 3.7 (− 6.1 to − 1.3)

NA 0.3 (0.2 to 0.5)

.003

b.001

CI, confidence interval; DFDs, disability-free days; NA, not applicable; PFDs, pain-free days; QALY, Quality-adjusted life years; SMT, spinal manipulation therapy. a This table does not include study costs of study treatment because these are nonstochastic and additive and could bias the multiplicative log model if included. There was no variability in study treatment costs. Note that study treatment costs are added into Table 5.

disability (P = .01) were predictors of higher annual treatment costs, whereas higher baseline QALY scores (P b .01) predicted lower treatment costs. In the model that also includes costs of lost productivity, college education (P = .02) and previous experience with massage (P = .04) were predictive of higher costs, whereas higher baseline QALY scores (P b .01) were predictive of lower costs. Costs in the 4 weeks preceding the baseline were strong predictors of higher costs in both these models (P b .01).

Pain-Free Days and LBP-Related DFDs Table 4 also shows the regression models for LBPrelated PFDs and DFDs that patients experienced during the 1-year study period. Compared with the SMT 0 reference

group, patients benefited from 16.8 to 22.9 more PFDs and 12.3 to 19.4 more DFDs. The greatest benefits in PFDs and DFDs were for the SMT 12 group and were the only statistically significant results (P = .03 and .04, respectively). In addition, baseline health as measured by pain, disability, and the QALY score was strongly predictive of the number of PFDs and DFDs that patients experienced. A 1-point increase in baseline pain on a 100-point scale predicted a 2.7-point decrease in PFDs (P b .01), and a 1point increase in baseline disability predicted a 1.4-point decrease in DFDs (P b .01). Quality-adjusted life year scores are measured on a much narrower, approximately 1point scale. Each 0.01-point increase in the baseline QALY score predicted 1.07 and 1.05 additional PFDs and DFDs during the study period (P = .01 and P b .01, respectively).

305

306


Quality-Adjusted Life Years


EuroQol-5D–based QALY scores increased 4% to 6% during the 1-year study period to 0.81 for SMT 0, 0.80 for SMT 6, 0.83 for SMT 12, and 0.81 for SMT 18. A regression analysis similar to the ones for PFDs/DFDs above found that relative to SMT 0, each dose of SMT yielded an additional 0.00 to 0.01 QALYs. None of the differences either between study groups or across time within study groups were statistically significant.

costs but somewhat more favorable to the highest dose of SMT for costs of treatment plus lost productivity (Table 3). Post hoc power analysis of costs of lost productivity and treatments outside the study protocol showed that our trial had 80% power to detect cost increases of 145% for SMT 6, 101% for SMT 12, and 107% for SMT 18 relative to the reference SMT 0 group. Our trial had 80% power to detect relative cost decreases compared with SMT 0 of 44%, 53%, and 50% for the 3 groups, respectively.

Incremental Effects of SMT Dose and Patient Characteristics on Adjusted Costs

DISCUSSION

The costs related to the 3 patient profiles are reported in Table 5. They include the cost of study-provided care, the cost of outside care, and, where indicated, the cost of lost productivity. The effects of treatment groups and changes in selected baseline characteristics are presented for each profile. For profile M, doses of SMT groups 6, 12, and 18 predicted increases of $317, $123, and $345 in annual treatment costs, respectively, when compared with no SMT. When lost productivity was included, the doses predicted increases of $734, $94, and $476. Compared with the SMT 0 group, the treatment cost of an additional PFD was $19, $5, and $19 and the treatment cost of an additional DFD was $25, $6, and $28 for SMT groups 6, 12, and 18, respectively. When treatment and lost productivity were considered, the cost of an additional PFD was $44, $4, and $26 and the cost of an additional DFD was $57, $5, and $39 for the 3 dose groups, respectively. Treatment and lost productivity costs were less responsive to SMT dose for the lower cost patient in profile L. Baseline patient characteristics such as smoking, age, and disability also had smaller effects on costs for such patients when compared with effects on profile M. For profile H, doses of SMT groups 6, 12, and 18 predicted increases of $1217, −$271, and $39 in annual treatment costs, respectively when compared with SMT 0. When the cost of lost productivity was included, the increases were predicted to be $2777, −$277, and $710. Compared with the SMT 0 group, the treatment cost of an additional PFD was $72, −$12, and $2, and the treatment cost of an additional DFD was $95, −$14, and $3 for SMT groups 6, 12, and 18 respectively. When treatment and lost productivity were considered, the cost of an additional PFD was $165, −$12, and $39, and the cost of an additional DFD was $217, −$14, and $58 for the 3 dose groups, respectively.

Sensitivity Analysis Sensitivity analyses that excluded weeks not covered by patient reports yielded results that were very similar to those obtained from analysis of annualized cost data for treatment

This study shows that manipulation can have a modicum of benefit in LBP-related pain and disability without significantly increasing treatment or societal costs. In the main results paper reporting pain and disability on 100point scales, Haas et al 24 found an 8.6-point advantage in pain (P = .002) and 7.5-point advantage in disability (P = .011) for SMT 12 compared with the control at 12 weeks. In this study, we found that this benefit may occur without significant increases in costs of treatment or lost productivity. We also found a modest incremental benefit of 23 PFDs and 19 DFDs from 12 spinal manipulation treatments relative to a no-manipulation control. The incremental benefits from the other doses examined showed an increase in PFDs and DFDs, but not with statistical significance. Spinal manipulative therapy dose did not have a statistically significant effect on either QALYs, treatment costs, or the combined cost of treatment and lost productivity. Finding no statistically significant QALY differences either between study groups or across time within study groups is consistent with several prior studies suggesting that QALY scores are insensitive to small or moderate changes in health. 32,51,52 It is not uncommon for studies of chronic conditions, especially chronic pain to use other measures of patient benefit in addition to QALYs. For example, Niemisto et al 53 used measures including VAS pain, self-rated disability, health-related quality of life, and patient satisfaction. A number of studies have previously examined the benefits and cost of SMT in patients with LBP. 50,53–58 The estimated benefits and costs have varied between studies, and part of the difference in results may arise because the studies examined different or unspecified doses of SMT. Our results indicate that incremental benefits vary with SMT dose with SMT 12 yielding the greatest benefit. Our analysis of the effect of the SMT dose on costs should be interpreted with caution because of the large variability in costs (low power). Substantially larger sample sizes appear to be required in trials of SMT in patients with chronic LBP in order to detect differences in costs of treatment and, especially, lost productivity between patients receiving different doses of SMT.

Incremental Cost ($) of Dose Change to

Patient type

Group

Predicted mean costs ($)

Incremental cost per PFD compared with SMT 0 ($)

Annual cost of treatment 596 NA Profile M a SMT 0 Profile M a SMT 6 914 19 Profile M a SMT 12 720 5 Profile M a SMT 18 941 19 61 NA Profile L b SMT 0 Profile L b SMT 6 215 9 Profile L b SMT 12 255 9 Profile L b SMT 18 461 22 3551 NA Profile H c SMT 0 Profile H c SMT 6 4768 72 Profile H c SMT 12 3280 − 12 2 Profile H c SMT 18 3590 Annual cost of treatment and lost productivity Profile M a SMT 0 3492 NA Profile M a SMT 6 4226 44 4 Profile M a SMT 12 3586 Profile M a SMT 18 3968 26 Profile L b SMT 0 1637 NA Profile L b SMT 6 2053 25 7 Profile L b SMT 12 1789 Profile L b SMT 18 2076 24 Profile H c SMT 0 15 422 NA 18 200 165 Profile H c SMT 6 Profile H c SMT 12 15 145 − 12 Profile H c SMT 18 16 133 39

Incremental cost per DFD compared with SMT 0 ($)

SMT 6

Incremental Costs ($) due to the Following: $30 (1 SD) increase in treatment costs in the 14 d preceding enrollment

$600 (1 SD) increase in treatment and lost productivity costs in the 14 d preceding enrollment

SMT 12

SMT 18 345 28 222 NA 401 246 206 NA 39 − 1178 310 NA

123 161 107 110 12 16 11 11 733 956 635 657

34 44 29 30 3 4 3 3 202 263 175 181

192 251 167 173 20 25 17 18 1147 1496 994 1028

119 156 104 107 47 61 41 42 487 635 422 436

NA NA NA NA NA NA NA NA NA NA NA NA

476 − 258 381 NA 439 23 287 NA 710 − 2067 988 NA

536 627 519 546 251 294 243 256 2366 2771 2292 2412

764 894 740 779 358 419 347 365 3373 3951 3268 3439

1601 1875 1551 1632 750 879 727 765 7071 8282 6851 7210

NA NA NA NA NA NA NA NA NA NA NA NA

825 966 799 841 387 453 375 394 3643 4267 3530 3715

NA 25 6 28 NA 12 10 33 NA 95 − 14 3

317 NA

123 − 194 NA

154 NA

195 41 NA

1217 NA

− 271 − 1488 NA

NA 57 5 39 NA 33 8 36 NA 217 − 14 58

734 NA

94 − 640 NA

416 NA

152 − 264 NA

2777 NA

− 277 − 3055 NA

Smoking

14-y (1 SD) increase in age

22-point (1 SD) increase in baseline disability score


NA, not applicable: retransformed profiles include additive cost of study treatment. a Profile M (patient characteristics reflective of the mean)—41-year-old, college-educated, non-white or Hispanic, unmarried, nonsmoking woman with no previous massage or chiropractic treatment, a differential treatment confidence score of 0, and baseline pain, disability, and QALY scores of 51, 45, and 0.76, respectively. She had treatment costs of $35 and treatment plus lost productivity costs of $600 in the 4 weeks preceding study enrollment. b Profile L (variables 1 SD away from mean predicted to lower costs)—27-year-old, non-white or Hispanic, unmarried, low-income, nonsmoking man with no previous massage or chiropractic treatment, a differential treatment confidence score of 0, and baseline pain, disability, and QALY scores of 60, 30, and 0.85, respectively. He had treatment costs of $5 and treatment plus lost productivity costs of $200 in the 4 weeks preceding study enrollment. c Profile H (variables 1 SD away mean predicted to raise costs)—55-year-old, college-educated, white non-Hispanic, married, nonsmoking woman with prior massage and chiropractic treatment, a differential treatment confidence score of 1, and baseline pain, disability, and QALY scores of 40, 55, and 0.65, respectively. She had treatment costs of $55 and treatment plus lost productivity costs of $1200 in the 4 weeks preceding study enrollment.


Table 5. Predicted Mean Costs and Incremental Effects of Predictors

307

308


The modest size of the incremental benefit from SMT 12 is not unusual for studies of interventions in patients with chronic pain. In their analysis of the cost of manipulation, stabilizing exercises, and physician consultation in comparison with physician consultation alone, Niemisto et al 53 found mean intervention group improvements relative to control of 6.3 points in pain and an improvement of 2.3 points in disability after 12 months when both pain and disability were measured on scales of 0 to 100. In a study of a collaborative intervention to treat chronic pain among veterans, the intervention yielded 16 additional pain DFDs relative to control. 32 Our results confirm previous findings that the cost of lost productivity due to LBP is often much higher than the cost of treatment. A systematic review of LBP cost of illness studies lists 8 studies that examined both direct (mostly treatment) and indirect (mostly lost productivity) costs. 7 In 6 of these studies, direct costs comprised less that 20% of combined direct and indirect costs. In our study, cost of productivity lost due to LBP is between approximately 2.5and 10-fold higher than the cost of treatment. Kominski et al 58 found the baseline characteristics of patients with LBP including age, education, and baseline health to be important predictors of treatment costs. Our cost analyses reemphasize the importance of similar baseline patient characteristics, especially baseline pain and disability, as important predictors of treatment costs and lost productivity. In a result consistent with a previous analysis of treatment costs among patients with chronic pain, our results show that costs in the period preceding enrollment in the trial can also be predictive of the costs of treatment and lost productivity. 32 A recent study on patients with back and neck problems found that CAM users had lower costs than did non–CAM users. 9 Our results suggest a potentially more complex relationship for LBP treated by doctors of chiropractic. For patients with characteristics predictive of high costs (profile H, Table 5), costs of treatment obtained outside the study were much larger than the costs of study-provided treatment, and costs of lost productivity were even larger. Furthermore, costs may increase with SMT for patients whose baseline characteristics predict low costs (profile L) in contrast to the decrease with some SMT doses for patients whose baseline characteristics predict high costs (profile H). This highlights the disproportionate share of the economic burden of LBP that patients similar to our profile H bear in terms of treatment and lost productivity. Larger trials are necessary to clarify the actual causal relationships. Although there are no published standardized measures of the individual financial consequences of pain, 59 there are measures of meaningful clinical improvement to pain and disability. Retrospective reports of interference with daily activities show internal consistency and convergent validity 59,60 and 11-point numerical rating scales provide sufficient levels of discrimination with similar sensitivity to


101-point scales. 59,61 Using first-return-to-work as an indicator of recovery is challenging because it is frequently followed by recurrences of work absence. 62 It is difficult to measure work disability outside the usual measures of absence from/presence at work and the number of days of work absence. New measures such as the clinical return-towork-rule and testing at-work disability may be of interest in future research. 59,63,64

Limitations The selected participants for this study were not on disability, nor were they seeking disability claims. In addition, potential participants with confounding conditions were excluded. Our sample size was too small to sufficiently capture participants who would eventually elect to get surgery, and the one patient who did so was removed as an outlier. Therefore, this study population best generalizes to a subset of the population missing work. In addition, because more than 70% of the participants had previous experience with SMT and more than 40% had experience with massage, the expectations of the patients were balanced at baseline. An analysis of the doctor-patient encounter showed that expectations were balanced across groups and had no notable effect on follow-up LBP. 65 Still, caution in interpretation of the results is advised. In this study, patients' mean baseline QALY score was 0.75, which is higher than that observed in most other LBP trials and may limit direct comparisons of results and may present ceiling effects. In their review of studies of interventions for LBP, Dagenais et al 66 compiled a list of baseline QALY scores in the reviewed studies (their Table 5) that ranged from 0.35 to 0.71; improvements in QALY scores reported by the studies tended to be smaller in groups with higher baseline scores. Similarly, the existence of a variety of payment systems and costing methodologies complicates direct comparisons of cost results from different studies. Our use of Medicare payments to estimate treatment costs may have contributed to our finding of low treatment costs relative to studies that have used charges, prices paid by private insurers, or other methods. Mean annual treatment costs of study-provided and outside care across the different SMT doses ranged from $315 to $897. Although these costs are broadly comparable with those reported by Herman et al 67 for the cost of naturopathic care for LBP, other studies have reported substantially higher costs, especially if treatment involved surgery. Martin et al 68 found through analysis of a nationally representative sample that patients with back and neck problems had mean costs of $6096 in 2005. The minimum cost of lumbar surgery has been estimated at $15 000. 69 Using patient reports of health at specific time points or during specific periods to estimate health in intervening periods is common in cost analyses, and we use this approach to compute DFDs for the complete 1-year


duration of the study. 31,66,70 However, it remains possible that this extrapolation may introduce an unknown bias into our estimates. We similarly use patient reports of health care use and activity restriction in 4-week periods preceding the 6, 12, 18, 24, 39, and 52-week time points to estimate annual treatment and lost productivity costs. The possibility of conservative bias due to extrapolation cannot be ruled out. Our estimates of the cost of productivity lost due to LBP are also conservative. Using mean as opposed to median wages to estimate the cost of each hour of productivity lost would have increased these costs by approximately 30%. Attributing a full workday (8 hours) of lost productivity to each day when a patient was kept from usual activities due to LBP would have compounded the increase by another third. The cost analyses, which yield results slightly more favorable to doses SMT 12 and SMT 18, are not statistically significant.

CONCLUSION Spinal manipulative therapy showed a modicum of benefit in LBP and LBP-related disability without increasing costs from a societal prospective (treatment costs and productivity together). Inference regarding cost is preliminary because of low statistical power. The SMT 12 group had the most PFDs and DFDs compared with SMT 0. The SMT 12 group was also favored when looking at predicted mean costs, but the incremental cost of SMT interventions

Practical Applications • The greatest benefits in PFDs and DFDs, experienced by participants over the entire year, were observed in those receiving 12 manipulation visits. • No differences were observed comparing the 3 dose levels to control, and no trends across time points or dose can be inferred given the variability of the estimates of treatment alone including outside care, as well as including the cost of lost productivity. • Higher annual treatment costs, in the model that also includes annual costs of lost productivity, increased with more college education, previous experience with massage, and lower baseline QALY scores. Costs in the 4 weeks preceding the baseline were also strongly associated with higher annual costs. • Treatment and lost productivity costs were most responsive to SMT dose in the highest cost patient profile, especially when lost productivity was included.


depended on the characteristics of patients being treated. Specifically, for patients who were likely to have high costs, the SMT 12 intervention was cost decreasing (but not statistically significant). This report adds to the evidence that a dose of 12 SMT treatments appears to be reasonable for use in a comparative effectiveness trial.

FUNDING SOURCES AND POTENTIAL CONFLICTS OF INTEREST No funding sources or conflicts of interest were reported for this study.

CONTRIBUTORSHIP INFORMATION Concept development (provided idea for the research): MH, RS. Design (planned the methods to generate the results): MH, RS. Supervision (provided oversight, responsible for organization and implementation, writing of the manuscript): DV, RS, MH. Data collection/processing (responsible for experiments, patient management, organization, or reporting data): MH, DV. Analysis/interpretation (responsible for statistical analysis, evaluation, and presentation of the results): RS, MH, DV. Literature search (performed the literature search): RS, DV, MH Writing (responsible for writing a substantive part of the manuscript): DV, RS, MH. Critical review (revised manuscript for intellectual content, this does not relate to spelling and grammar checking): MH, RS, DV.

REFERENCES 1. Andersson GBJ. Epidemiological features of chronic lowback pain. Lancet 1999;354:581-5. 2. Deyo RA, Mirza SK, Martin BI. Back pain prevalence and visit rates: estimates from U.S. national surveys, 2002. Spine 2006;31:2724-7. 3. Woolf AD, Pfleger B. Burden of major musculoskeletal conditions. Bull World Health Organ 2003;81:646-56. 4. Hart LG, Deyo RA, Cherkin DC. Physician office visits for low back pain. Spine 1995;20:11-9. 5. Gore M, Sadosky A, Stacey BR, Tai KS, Leslie D. The burden of chronic low back pain: clinical comorbidities, treatment patterns, and health care costs in usual care settings. Spine 2012;37:E668-77. 6. Lind BK, Lafferty WE, Tyree PT, Sherman KJ, Deyo RA, Cherkin DC. The role of alternative medical providers for the outpatient treatment of insured patients with back pain. Spine 2005;30:1454-9.

309

310


7. Dagenais S, Caro J, Haldeman S. A systematic review of low back pain cost of illness studies in the United States and internationally. Spine J 2008;8:8-20. 8. Kessler RC, Davis RB, Foster DF, et al. Long-term trends in the use of complementary and alternative medical therapies in the United States. Ann Intern Med 2001;135:262-8. 9. Martin BI, Gerkovich MM, Deyo RA, et al. The association of complementary and alternative medicine use and health care expenditures for back and neck problems. Med Care 2012;50: 1029-36. 10. Haas M, Sharma R, Stano M. Cost-effectiveness of medical and chiropractic care for acute and chronic low back pain. J Manipulative Physiol Ther 2005;28:555-63. 11. Sharma R, Haas M, Stano M, Spegman A, Gehring R. Determinants of costs and pain improvement for medical and chiropractic care of low back pain. J Manipulative Physiol Ther 2009;32:252-61. 12. Goetzel RZ, Hawkins K, Ozminkowski RJ, Wang S. The health and productivity cost burden of the “top 10” physical and mental health conditions affecting six large U.S. employers in 1999. J Occup Environ Med 2003;45:5-14. 13. Smith M, Davis MA, Stano M, Whedon JM. Aging baby boomers and the rising cost of chronic back pain: secular trend analysis of longitudinal medical expenditures panel survey data for years 2000 to 2007. J Manipulative Physiol Ther 2013;36:2-11. 14. Hong J, Reed C, Novick D, Happich M. Costs associated with treatment of chronic low back pain: an analysis of the UK general practice research database. Spine 2013;38:75-82. 15. Nimgade A, McNeely E, Milton D, Celona J. Increased expenditures for other health conditions after an incident of low back pain. Spine 2010;35:769-77. 16. Smith M. Identifying episodes of back pain using medical expenditures panel survey data: patient experience, use of services, and chronicity. J Manipulative Physiol Ther 2010;33:562-75. 17. Liliedahl RL, Finch MD, Axene DV, Goertz CM. Cost of care for common back pain conditions initiated with chiropractic doctor vs medical doctor/doctor of osteopathy as first physician: experience of one Tennessee-based general health insurer. J Manipulative Physiol Ther 2010;33:640-3. 18. Weigel PA, Hockenberry JM, Bentler SE, Kaskie B, Wolinsky FD. Chiropractic episodes and the co-occurrence of chiropractic and health services use among older Medicare beneficiaries. J Manipulative Physiol Ther 2012;35:168-75. 19. Haas M, Spegman A, Peterson D, Aickin M, Vavrek D. Dose response and efficacy of spinal manipulation for chronic cervicogenic headache: a pilot randomized controlled trial. Spine J 2010;10:117-28. 20. Haas M, Groupp E, Aickin M, et al. Dose-response for chiropractic care of chronic cervicogenic headache and associated neck pain: a randomized pilot study. J Manipulative Physiol Ther 2004;27:547-53. 21. Haas M, Groupp E, Kraemer DF. Dose-response for chiropractic care of chronic low back pain. Spine J 2004;4:574-83. 22. Bronfort G, Haas M, Evans R, Leininger B, Triano J. Effectiveness of manual therapies: the UK evidence report. Chiropr Osteopat 2010;18:3. 23. Rubinstein SM, van Middelkoop M, Assendelft WJ, de Boer MR, van Tulder MW. Spinal manipulative therapy for chronic low-back pain: an update of a Cochrane review. Spine 2011; 36:E825-46. 24. Haas M, Vavrek D, Peterson D, Polissar N, Neradilek MB. Doseresponse and efficacy of spinal manipulation for care of chronic low back pain: a randomized controlled trial. Spine J 2013 [Epub ahead of print]. 25. Anderson JAD. Problems of classification of low-back pain. Rheumatol Rehabil 1977;16:34-6.


26. DiFabio RP, Mackey G, Holte JB. Physical therapy outcomes for patients receiving workers' compensation following treatment for herniated lumbar disc and mechanical low back pain syndrome. J Orthop Sports Phys Ther 1996;23:180-7. 27. Shekelle P, Adams A, Chassin M, Hurwitz E, Brook R. Spinal manipulation for low-back pain. Ann Intern Med 1992;117: 590-8. 28. Peterson DH, Bergmann TF. Chiropractic technique: principles and practice. St. Louis: Mosby; 2002. 29. Furlan AD, Brosseau L, Imamura M, Irvin E. Massage for low back pain: a systematic review within the framework of the Cochrane Collaboration Back Review Group. Spine 2002;27: 1896-910. 30. Cherkin DC, Deyo RA, Sherman KJ, et al. Characteristics of licensed acupuncturists, chiropractors, massage therapists and naturopathic physicians. J Am Board Fam Pract 2002;15:463-72. 31. Lave JR, Frank RG, Schulberg HC, Kamlet MS. Costeffectiveness of treatments for major depression in primary care practice. Arch Gen Psychiatry 1998;55:645-51. 32. Dickinson KC, Sharma R, Duckart JP, Corson K, Gerrity MS, Dobscha SK. VA healthcare costs of a collaborative intervention for chronic pain in primary care. Med Care 2010;48:38-44. 33. Underwood MR, Barnett AG, Vickers MR. Evaluation of two time-specific back pain outcome measures. Spine 1999;24: 1104-12. 34. Von Korff M, Ormel J, Keefe FJ, Dworkin SF. Grading the severity of chronic pain. Pain 1992;50:133-49. 35. Airaksinen O, Brox JI, Cedraschi C, et al. Chapter 4. European guidelines for the management of chronic nonspecific low back pain. Eur Spine J 2006;15(Suppl 2):S192-300. 36. The EuroQol Group. EuroQol—a new facility for the measurement of health-related quality of life. Health Policy 1990;16:199-208. 37. Brooks R. EuroQol: the current state of play. Health Policy 1996;37:53-72. 38. Ritter PL, Stewart AL, Kaymaz H, Sobel DS, Block DA, Lorig KR. Self-reports of health care utilization compared to provider records. J Clin Epidemiol 2001;54:136-41. 39. Medicare's 2009 national non-facility (i.e., non-hospital) payments. Available from: www.cms.gov/apps/physician-feeschedule/ [Last accessed on:8/10/2012]. 40. Folland S, Goodman AC, Stano M. The economics of health and health care. 6th ed. Boston: Prentice Hall; 2010 [601 pp.]. 41. Hospital Outpatient Prospective Payment System Limited Data Set. Available from: http://www.cms.gov/researchstatistics-data-and-systems/files-for-order/LimitedDataSets/ HospitalOPPS.html [Last accessed on:01/11/2013]. 42. Dionne CE, Von Korff M, Koepsell TD, Deyo RA, Barlow WE, Checkoway H. A comparison of pain, functional limitations, and work status indices as outcome measures in back pain research. Spine 1999;24:2339-45. 43. Stewart WF, Ricci JA, Chee E, Morganstein D, Lipton R. Lost productive time and cost due to common pain conditions in the US workforce. JAMA 2003;290:2443-54. 44. Mattke S, Balakrishnan A, Bergamo G, Newberry SJ. A review of methods to measure health-related productivity loss. Am J Manag Care 2007;13:211-7. 45. Mitchell RJ, Bates P. Measuring health-related productivity loss. Popul Health Manag 2011;14:93-8. 46. Bureau of Labor Statistics, U.S. Department of Labor. OES highlights: an overview of u.s. occupational employment and wages in 2009. June 2010. Accessed 02/05/2013. 47. Diehr P, Yanez D, Ash A, Hornbrook M, Lin DY. Methods for analyzing health care utilization and costs. Annu Rev Public Health 1999;20:125-44.


48. Duan N. Smearing estimate: a nonparametric retransformation method. J Am Stat Assoc 1983;78:606-10. 49. Judge GG, Hill RC, Griffiths WE, Lutkepohl H, Lee T. Chapter 20. Introduction to the theory and practice of econometrics. 2nd ed. John Wiley & Sons; 1988. 50. Sharma R, Haas M, Stano M. Patient attitudes, insurance, and other determinants of self-referral to medical and chiropractic physicians. Am J Public Health 2003;93:2111-7. 51. Jenkinson C, Gray A, Doll H, Lawrence K, Keoghane S, Layte R. Evaluation of index and profile measures of health status in a randomized controlled trial. Med Care 1997;35: 1109-18. 52. Harper R, Brazier JE, Waterhouse JC, Walters SJ, Jones NM, Howard P. Comparison of outcome measures for patients with chronic obstructive pulmonary disease (COPD) in an outpatient setting. Thorax 1997;52:879-87. 53. Niemisto L, Rissanen P, Sarna S, Lahtinen-Suopanki T, Lindgren KA, Hurri H. Cost-effectiveness of combined manipulation, stabilizing exercises, and physician consultation compared to physician consultation alone for chronic low back pain: a prospective randomized trial with 2-year followup. Spine 2005;30:1109-15. 54. McMorland G, Suter E. Chiropractic management of mechanical neck and low-back pain: a retrospective, outcome-based analysis. J Manipulative Physiol Ther 2000;23:307-11. 55. Niemisto L, Lahtinen-Suopanki T, Rissanen P, Lindgren K, Sarna S, Hurri H. A randomized trial of combined manipulation, stabilizing exercises, and physician consultation compared to physician consultation alone for chronic low back pain. Spine 2003;28:2185-91. 56. Uk BEAM Trial Team. United Kingdom Back Pain Exercise and Manipulation (UK BEAM) randomised trial: effectiveness of physical treatments for back pain in primary care. BMJ 2004;329:1377-81. 57. Uk BEAM Trial Team. United Kingdom Back Pain Exercise and Manipulation (UK BEAM) randomised trial: cost effectiveness of physical treatments for back pain in primary care. BMJ 2004;329:1381-5. 58. Kominski GF, Heslin KC, Morgenstern H, Hurwitz EL, Harber PI. Economic evaluation of four treatments for low-


59. 60. 61. 62. 63. 64. 65.

66.

67. 68. 69. 70.

back pain: results from a randomized controlled trial. Med Care 2005;43:428-35. Croft P, Blyth F, van der Windt D. Chronic pain epidemiology: from aetiology to public health. Oxford: Oxford University Press; 2010 [365 pp.]. Turk DC, Melzack R. Handbook of pain assessment. 2nd ed. New York: Guilford Press; 2001 [760 pp.]. Jensen MP, Turner JA, Romano JM. What is the maximum number of levels needed in pain intensity measurement? Pain 1994;58:387-92. Baldwin ML, Johnson WG, Butler RJ. The error of using returns-to-work to measure the outcomes of health care. Am J Ind Med 1996;29:632-41. Beaton DE, Kennedy CA. Beyond return to work: testing a measure of at-work disability in workers with musculoskeletal pain. Qual Life Res 2005;14:1869-79. Dionne CE, Bourbonnais R, Fremont P, Rossignol M, Stock SR, Larocque I. A clinical return-to-work rule for patients with back pain. CMAJ 2005;172:1559-67. Haas M, Vavrek D, Neradilek MB, Polissar N. A path analysis of the effects of the doctor-patient encounter and expectancy in an open-label randomized trial of spinal manipulation for the care of low back pain. BMC Complement Altern Med 2014;14:16 [6882-14-16]. Dagenais S, Roffey DM, Wai EK, Haldeman S, Caro J. Can cost utility evaluations inform decision making about interventions for low back pain? Spine J 2009;9: 944-57. Herman PM, Szczurko O, Cooley K, Mills E. Costeffectiveness of naturopathic care for chronic low back pain. Altern Ther Health Med 2008;14:32-8. Martin BI, Deyo RA, Mirza SK, et al. Expenditures and health status among adults with back and neck problems. JAMA 2008;299:656-64. Turk DC. Clinical effectiveness and cost-effectiveness of treatments for patients with chronic pain. Clin J Pain 2002;18:355-65. Dakin H, Gray A, Fitzpatrick R, Maclennan G, Murray D, KAT Trial Group. Rationing of total knee replacement: a cost-effectiveness analysis on a large trial data set. BMJ Open 2012;2:e000332 [2011-000332. Print 2012].

311

Spinal manipulative therapy for low back pain.

Spinal Manipulative Therapy for Chronic Lower Back Pain in Older Veterans: A Prospective, Randomized, Placebo-Controlled Trial.

Evaluation of a modified clinical prediction rule for use with spinal manipulative therapy in patients with chronic low back pain: a randomized clinical trial.

Spinal manipulative therapy for low back pain-time for an update.

A randomized controlled trial of hypnosis compared with biofeedback for adults with chronic low back pain.

Prediction of pain outcomes in a randomized controlled trial of dose-response of spinal manipulation for the care of chronic low back pain.

Effectiveness of Chinese massage therapy (Tui Na) for chronic low back pain: study protocol for a randomized controlled trial.

Standardized versus Individualized Acupuncture for Chronic Low Back Pain: A Randomized Controlled Trial.

Fascial Manipulation® for chronic aspecific low back pain: a single blinded randomized controlled trial.

Fascial Manipulation® for chronic aspecific low back pain: a single blinded randomized controlled trial.

Paraspinous Lidocaine Injection for Chronic Nonspecific Low Back Pain: A Randomized Controlled Clinical Trial.

Pain education to prevent chronic low back pain: a study protocol for a randomised controlled trial.

Spinal manipulative therapy-specific changes in pain sensitivity in individuals with low back pain (NCT01168999).

Chiropractic spinal manipulative therapy for cervicogenic headache: a single-blinded, placebo, randomized controlled trial.

Qigong versus exercise therapy for chronic low back pain in adults--a randomized controlled non-inferiority trial.

Acupuncture for low back pain due to spondylolisthesis: study protocol for a randomized controlled pilot trial.

Effects of unstable shoes on chronic low back pain in health professionals: a randomized controlled trial.

Open-label placebo treatment in chronic low back pain: a randomized controlled trial.

Rational and design of an individual participant data meta-analysis of spinal manipulative therapy for chronic low back pain-a protocol.

Spinal manipulation and exercise for low back pain in adolescents: study protocol for a randomized controlled trial.

A randomized, placebo-controlled trial of patient education for acute low back pain (PREVENT Trial): statistical analysis plan.

Randomized controlled trial on the effectiveness of cognitive behavior group therapy in chronic back pain patients.

Erratum to: Spinal manipulative therapy, Graston technique® and placebo for non-specific thoracic spine pain: a randomised controlled trial.

Is exposure in vivo cost-effective for chronic low back pain? A trial-based economic evaluation.